You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@superset.apache.org by ru...@apache.org on 2026/05/14 16:39:41 UTC

(superset) 01/01: test(datasets): regression coverage for #16141 (export with same table name, different schemas)

This is an automated email from the ASF dual-hosted git repository.

rusackas pushed a commit to branch test/16141-dataset-export-same-name-schema
in repository https://gitbox.apache.org/repos/asf/superset.git

commit 233eb49bb27f392496db327015530bdeb1af1dbf
Author: Superset Dev <de...@superset.apache.org>
AuthorDate: Thu May 14 09:39:05 2026 -0700

    test(datasets): regression coverage for #16141 (export with same table name, different schemas)
    
    Issue #16141 (Aug 2021) reported that the dataset export API returned
    only one file when exporting two datasets that shared a table name but
    lived in different schemas (e.g. prod.users + dev.users). The current
    ExportDatasetsCommand disambiguates the export filename with the
    dataset id (`<table_name>_<id>.yaml`), so the collision should no
    longer occur — but no test pinned the behavior.
    
    This is a TDD-first PR: the test asserts the *fixed* behavior. If CI
    green, the bug is provably gone and #16141 can be closed; if CI red,
    the test documents exactly what still needs to be fixed for whoever
    picks up the work.
    
    Closes #16141
    
    Co-Authored-By: Claude Opus 4.7 <no...@anthropic.com>
---
 tests/unit_tests/datasets/commands/export_test.py | 51 +++++++++++++++++++++++
 1 file changed, 51 insertions(+)

diff --git a/tests/unit_tests/datasets/commands/export_test.py b/tests/unit_tests/datasets/commands/export_test.py
index a449b764084..df02e0d29b5 100644
--- a/tests/unit_tests/datasets/commands/export_test.py
+++ b/tests/unit_tests/datasets/commands/export_test.py
@@ -304,3 +304,54 @@ version: 1.0.0
 """,
         ),
     ]
+
+
+def test_export_two_datasets_same_table_name_different_schema(
+    session: Session,
+) -> None:
+    """
+    Regression coverage for GitHub issue #16141.
+
+    Exporting two datasets that share a `table_name` but live in
+    different schemas (e.g. prod.users + dev.users) must produce two
+    distinct entries in the export. Historically the pair could collide
+    onto a single filename — the export filename is now disambiguated by
+    dataset id, so this test pins that behavior so it can't silently
+    regress.
+    """
+    from superset.commands.dataset.export import ExportDatasetsCommand
+    from superset.connectors.sqla.models import SqlaTable
+    from superset.models.core import Database
+
+    engine = db.session.get_bind()
+    SqlaTable.metadata.create_all(engine)  # pylint: disable=no-member
+
+    database = Database(database_name="my_database", sqlalchemy_uri="sqlite://")
+    db.session.add(database)
+    db.session.flush()
+
+    prod = SqlaTable(table_name="users", schema="prod", database=database)
+    dev = SqlaTable(table_name="users", schema="dev", database=database)
+    db.session.add_all([prod, dev])
+    db.session.flush()
+
+    paths: list[str] = []
+    contents: list[str] = []
+    for ds in (prod, dev):
+        for path, content_fn in ExportDatasetsCommand._export(  # pylint: disable=protected-access
+            ds, export_related=False
+        ):
+            paths.append(path)
+            contents.append(content_fn())
+
+    # Both datasets must produce distinct export paths — no collision.
+    assert len(paths) == len(set(paths)), (
+        f"Export filenames collided for same-table-name datasets: {paths}"
+    )
+
+    # And both YAML payloads must reflect their own schema, not be
+    # silently merged or overwritten.
+    schemas_in_yaml = {c.split("schema:")[1].splitlines()[0].strip() for c in contents}
+    assert schemas_in_yaml == {"prod", "dev"}, (
+        f"Expected both prod and dev schemas in export, got {schemas_in_yaml}"
+    )